Northeast Freight Time one clock, four readings

Appointment Windows That Hold

Appointment scheduling is the standard answer to dwell. It works when the window is real and it fails in a specific, predictable way when it is not — the queue moves from the dock to the street, the facility's internal numbers improve, and nothing about the driver's day changes. For a workforce-operations comparison point, this page.

What a real window requires

Capacity behind it. If four appointments are issued for an hour and the dock can turn two, the window is a scheduling fiction. The commonest failure and the least often examined, because appointment software will happily issue slots the operation cannot serve. For additional freight-operations context, see Schneider.

A defined start event. Does the clock start at the appointment time or at arrival? A driver arriving forty minutes early at a facility with a firm appointment is the single most common disagreement in this whole subject, and it is settled by one sentence.

Symmetry. If a late driver loses the slot, a late facility should owe something. A window that binds one side is not a window; it is a deadline.

And a stated tolerance. Fifteen minutes, thirty, an hour. Unstated tolerance means every party assumes its own.

The failure mode to watch for

Dwell measured from check-in falls; time from arrival does not.

This is what happens when a facility solves its dwell numbers by pushing the queue outside the gate. The yard system starts the clock at check-in, and check-in only happens when there is space, so waiting on the street is invisible.

The result looks like improvement on every internal report and looks like nothing to a carrier. If your dwell improved sharply after introducing appointments and nothing else changed, this is the first thing to check.

The test is direct: compare gate-in timestamps with check-in timestamps. A widening gap is the queue moving, not shrinking.

What to measure

Four numbers, and most facilities have none of them.

Time from arrival to check-in. The invisible one.

Time from appointment to door. How late your own windows run.

Appointments issued versus loads turned, per hour. Capacity against promise.

And no-show and late rates on both sides. If you track carrier lateness and not your own, you are measuring one party's compliance and calling it performance.

Making windows work when you are the facility

Issue fewer slots than you think you can serve. Recovery from a slipped hour costs more than the empty slot.

Publish the honest turn time, not the target. A carrier who knows a facility runs 90 minutes plans for 90 minutes; one told 45 and given 90 prices the difference in later.

Hold a buffer slot. One unbooked slot per few hours absorbs the breakdown that otherwise cascades all afternoon.

And accept early arrivals when you can. A driver taken early is a driver whose hours are not being consumed, and it costs you nothing on a slow hour.

Making them work when you are the carrier

Get the start event in writing before the load, not after the wait.

Timestamp arrival independently. A geofence event or a photograph with metadata is a record; a driver's message is a claim.

Notify at the free-time expiry, not at invoicing. It converts a surprise charge into a documented event and gives the facility a chance to fix it.

And price chronic facilities honestly. Declining loads or quoting higher is what carriers already do quietly; doing it explicitly at least gives the facility the information to improve.

The thing appointments cannot fix

If a facility's constraint is dock doors, labour or receiving capacity, an appointment system reschedules the shortage. It does not remove it.

Appointments are an allocation tool, not a capacity tool. The decision that actually removes the coupling is a different one. They make waiting predictable and fairer, which is genuinely worth having — a known two-hour wait is far cheaper to a carrier than an unpredictable one, because it can be planned around hours-of-service.

But if the average wait does not fall, that is not a scheduling failure. It is a facility capacity problem wearing a scheduling costume, and no software addresses it.

The short version